Photo By Lori McFate | Drew Walter, deputy assistant secretary of defense for nuclear matters, in blue...... read more read more
The Blue Grass Army Depot is the only Department of Defense organization testing artificial intelligence-powered surveillance software to detect threats such as intruders, weapons, and abnormal behavior in real time.
For the past eight months, BGAD has tested software called Scylla (pronounced SKY-la) as an intruder and active-shooter detection program, using wide-area detection cameras and drones. Unlike traditional surveillance systems, Scylla doesn’t just record incidents but proactively identifies risks and alerts security personnel before they escalate.
“If you’re the security operator, do you think you could watch 15 cameras at one time … and pick out a gun at 1,000 feet? Scylla can,” said Chris Willoughby, BGAD’s electronic security systems manager. (Incidentally, BGAD has a lot more than 15 cameras.)
In tests, Scylla detected an “intruder” with a gun climbing a water tower from an existing camera a mile away. Later, a closed-circuit TV camera that was 660 feet away detected the armed “intruder” kneeling on the water tower’s catwalk, and the camera zoomed in for a closer look.
On Sept. 18, BGAD gave an overview and demonstration to show how Scylla works. Among the 50 people attending were Drew Walter, deputy assistant secretary of defense for nuclear matters, and representatives for U.S. Rep. Andy Barr, U.S. Sen. Mitch McConnell, and U.S. Sen. Rand Paul. Also in attendance was Chris Grollnek, one of the country’s foremost authorities on active shooters, and representatives from the Madison County and Berea Independent school districts.
“Leveraging technology while building on the human experience, SCYLLA's true neural network artificial intelligence is fundamentally changing the landscape of how we prevent and respond,” Grollnek said. “Once a bad actor is in motion, whether a terrorist threat (domestic or international), foreign force, or violent criminal attack, this comprehensive initiative that Blue Grass Army Depot has created will prove critical in foiling an attack."
The demonstration showed how Scylla alerted security within seconds to two individuals carrying weapons. Facial recognition identified the men as BGAD personnel. Another demonstration showed three individuals breaching a security fence; a drone then followed their movements before they were stopped by BGAD security.
A third demonstration detected smoke coming from the bed of a blue pickup truck; the camera was about 700 feet away. When two men started a mock fight by swinging blows at each other, Scylla detected that behavior within seconds.
“It looks at video feeds to look for specific acts,” Willoughby said of Scylla.
The system won’t issue alerts upon detecting deer or wild turkeys – commons sights on the BGAD’s property – or swaying tree branches or holstered handguns.
The Army and DOD Physical Security Enterprise and Analysis Group selected the Scylla project submitted by BGAD as one of the top overall submissions in Fiscal Year 2024.
“We developed it, wrote it up, wrote the Conduct of Operations, submitted it, and they selected it and provided us funding,” Willoughby said.
And in terms of cost, Scylla is relatively inexpensive. In the same way that a Roku streaming stick connected to a TV’s HDMI port can make it a “smart TV” with your favorite streaming services, Scylla allows existing closed-circuit TV cameras to be “smart.”
It’s too early to say whether Scylla will be deployed at other military posts, but it shows promise in other applications, particularly for schools. Because it provides real-time alerts to security personnel, it allows them to respond quickly to potential threats rather than reacting to them after the fact.
By: Greg Kocher, Communications Specialist, Bechtel Parsons Blue Grass.
| Date Taken: | 10.08.2024 |
| Date Posted: | 10.08.2024 11:51 |
| Story ID: | 482745 |
| Location: | RICHMOND, KENTUCKY, US |
| Web Views: | 69 |
| Downloads: | 0 |
This work, A new tool for intruder detection: BGAD tests AI for enhanced security, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.
No keywords found.
